wiki/jogobot-red
1
0
Fork 0
Code Issues 4 Pull Requests Releases Wiki Activity

Apply changes to data structure

Browse Source 
See related ticket

Related Task: [https://fs.golderweb.de/index.php?do=details&task_id=94 FS#94]
This commit is contained in:
Jonathan Golder
2016-08-30 14:28:28 +02:00
parent f021f2ea60
commit 6149dcdb8b
4 changed files with 222 additions and 111 deletions
Download Patch File Download Diff File Expand all files Collapse all files

156
lib/redfam.py
View File

@@ -43,8 +43,8 @@ class RedFam:
Basic class for RedFams, containing the basic data structure
"""
def __init__( self, articlesList, beginning, ending=None, red_page_id=None,
status=0, fam_hash=None, heading=None ):
def __init__( self, articlesList, beginning, ending=None, redpageid=None,
status=None, famhash=None, heading=None ):
"""
Generates a new RedFam object
@@ -52,7 +52,7 @@ class RedFam:
@param beginning datetime Beginning date
@param ending datetime Ending date
@param red_page_id int MW pageid of containing RedPage
@param status int Status of RedFam
@param status str Status of RedFam
@param fam_hash str SHA1 hash of articlesList
@param heading str Original heading of RedFam (Link)
"""
@@ -61,20 +61,20 @@ class RedFam:
self.site = pywikibot.Site()
# Database interface
self._mysql = MysqlRedFam( fam_hash )
self._mysql = MysqlRedFam( famhash )
# Initial attribute values
self._articlesList = articlesList
self._beginning = beginning
self._ending = ending
self._red_page_id = red_page_id
self._status = status
self._fam_hash = fam_hash
self._redpageid = redpageid
self._status = self._parse_status(status)
self._famhash = famhash
self._heading = heading
# Calculates the sha1 hash over self._articlesList to
# rediscover known redundance families
self.calc_fam_hash()
self.calc_famhash()
def __repr__( self ):
"""
@@ -88,14 +88,14 @@ class RedFam:
", heading=" + repr( self._heading ) + \
", beginning=" + repr( self._beginning ) + \
", ending=" + repr( self._ending ) + \
", red_page_id=" + repr( self._red_page_id ) + \
", red_page_id=" + repr( self._redpageid ) + \
", status=" + repr( self._status ) + \
", fam_hash=" + repr( self._fam_hash ) + \
", fam_hash=" + repr( self._famhash ) + \
" )"
return __repr
def calc_fam_hash( self ):
def calc_famhash( self ):
"""
Calculates the SHA-1 hash for the articlesList of redundance family.
Since we don't need security SHA-1 is just fine.
@@ -106,35 +106,35 @@ class RedFam:
h = hashlib.sha1()
h.update( str( self._articlesList[:8] ).encode('utf-8') )
if self._fam_hash and h.hexdigest() != self._fam_hash:
raise RedFamHashError( self._fam_hash, h.hexdigest() )
if self._famhash and h.hexdigest() != self._famhash:
raise RedFamHashError( self._famhash, h.hexdigest() )
elif self._fam_hash:
elif self._famhash:
return
else:
self._fam_hash = h.hexdigest()
self._famhash = h.hexdigest()
def changed( self ):
"""
Checks wether anything has changed and maybe triggers db update
"""
# On archived red_fams do not delete possibly existing ending
if( not self._ending and self._status > 1 and
# On archived redfams do not delete possibly existing ending
if( not self._ending and "archived" in self._status and
self._mysql.data[ 'ending' ] ):
self._ending = self._mysql.data[ 'ending' ]
# Since status change means something has changed, update database
if( self._status != self._mysql.data[ 'status' ] or
if( self._raw_status != self._mysql.data[ 'status' ] or
self._beginning != self._mysql.data[ 'beginning' ] or
self._ending != self._mysql.data[ 'ending' ] or
self._red_page_id != self._mysql.data[ 'red_page_id' ] or
self._red_page_id != self._mysql.data[ 'redpageid' ] or
self._heading != self._mysql.data[ 'heading' ]):
self._mysql.update_fam( self._red_page_id, self._heading,
self._mysql.update_fam( self._redpageid, self._heading,
self._beginning, self._ending,
self._status )
self._raw_status() )
@classmethod
def flush_db_cache( cls ):
@@ -143,6 +143,61 @@ class RedFam:
"""
MysqlRedFam.flush()
def add_status(self, status):
"""
Adds a status specified by status, to status set
@param status Statusstring to add
@type status str
"""
self._status.add(status)
def remove_status(self, status, weak=True):
"""
Removes a status, specified by status from set. If weak is set to
False it will throw a KeyError when trying to remove a status not set.
@param status Statusstring to add
@type status str
@param weak Change behavior on missing status
@type bool
"""
if weak:
self._status.discard(status)
else:
self._status.remove(status)
def has_status(self, status):
"""
Returns True, if redfam has given status
@param status Statusstring to check
@type status str
@returns True if status is present else False
"""
if status in self._status:
return True
else:
return False
def _parse_status(self, raw_status ):
"""
Sets status based on comma separated list
@param raw_status Commaseparated string of stati (from DB)
@type raw_status str
"""
self._status = set( raw_status.strip().split(","))
def _raw_status( self ):
"""
Returns status as commaseparated string (to save in DB)
@returns Raw status string
@rtype str
"""
return ",".join( self._status )
class RedFamParser( RedFam ):
"""
@@ -165,15 +220,15 @@ class RedFamParser( RedFam ):
wurde gewünscht von:"
__done_notice2 = "{{Erledigt|"
def __init__( self, heading, red_page, red_page_archive,
def __init__( self, heading, redpage, redpagearchive,
beginning, ending=None ):
"""
Creates a RedFam object based on data collected while parsing red_pages
combined with possibly former known data from db
@param red_fam_heading str Wikitext heading of section
@param red_page page Pywikibot.page object
@param red_page_archive bool Is red_page an archive
@param redfam_heading str Wikitext heading of section
@param redpage page Pywikibot.page object
@param redpagearchive bool Is red_page an archive
@param beginning datetime Timestamp of beginning
str as strptime parseable string
@param ending datetime Timestamp of ending
@@ -181,9 +236,9 @@ class RedFamParser( RedFam ):
"""
# Set object attributes:
self._red_page_id = red_page._pageid
self._red_page_archive = red_page_archive
self._fam_hash = None
self._redpageid = redpage._pageid
self._redpagearchive = redpagearchive
self._famhash = None
# Method self.add_beginning sets self._beginning directly
self.add_beginning( beginning )
@@ -195,7 +250,7 @@ class RedFamParser( RedFam ):
# If no ending was provided set to None
self._ending = None
self._status = None
self._status = set()
# Parse the provided heading of redundance section
# to set self._articlesList
@@ -204,7 +259,7 @@ class RedFamParser( RedFam ):
# Calculates the sha1 hash over self._articlesList to
# rediscover known redundance families
self.calc_fam_hash()
self.calc_famhash()
# Open database connection, ask for data if existing,
# otherwise create entry
@@ -223,11 +278,11 @@ class RedFamParser( RedFam ):
# We need a connection to our mysqldb
self._mysql = MysqlRedFam( )
self._mysql.get_fam( self._fam_hash )
self._mysql.get_fam( self._famhash )
if not self._mysql.data:
self._mysql.add_fam( self._articlesList, self._heading,
self._red_page_id, self._beginning,
self._redpageid, self._beginning,
self._ending )
def heading_parser( self, heading ):
@@ -253,7 +308,7 @@ class RedFamParser( RedFam ):
# Catch sections with more then 8 articles, print error
if len( self._articlesList ) > 8:
# For repression in output we need to know the fam hash
self.calc_fam_hash()
self.calc_famhash()
jogobot.output(
( "\03{{lightred}}" +
@@ -317,21 +372,18 @@ class RedFamParser( RedFam ):
- 3 and greater status was set by worker script, do not change it
"""
# Do not change stati set by worker script etc.
if not self._mysql.data['status'] > 2:
# No ending, discussion is running:
# Sometimes archived discussions also have no detectable ending
if not self._ending and not self._red_page_archive:
self._status = 0
else:
if not self._red_page_archive:
self._status = 1
else:
self._status = 2
# No ending, discussion is running:
# Sometimes archived discussions also have no detectable ending
if not self._ending and not self._redpagearchive:
self.add_status("open")
else:
self._status = self._mysql.data[ 'status' ]
self.remove_status("open")
if not self._redpagearchive:
self.add_status("done")
else:
self.remove_status("done")
self.remove_status("open")
self.add_status("archived")
@classmethod
def is_section_redfam_cb( cls, heading ):
@@ -444,15 +496,15 @@ class RedFamWorker( RedFam ):
articlesList.append( mysql_data[ key ] )
super().__init__( articlesList, mysql_data[ 'beginning' ],
mysql_data[ 'ending' ], mysql_data[ 'red_page_id' ],
mysql_data[ 'status' ], mysql_data[ 'fam_hash' ],
mysql_data[ 'ending' ], mysql_data[ 'redpageid' ],
mysql_data[ 'status' ], mysql_data[ 'famhash' ],
mysql_data[ 'heading' ] )
self._mysql.data = mysql_data
# Get related RedPage-Information
self.redpageid = mysql_data[ 'page_id' ]
self.redpagetitle = mysql_data[ 'page_title' ]
self.redpageid = mysql_data[ 'pageid' ]
self.redpagetitle = mysql_data[ 'pagetitle' ]
# Make sure locale is set to 'de_DE.UTF-8' to prevent problems
# with wrong month abreviations in strptime
@@ -499,7 +551,7 @@ class RedFamWorker( RedFam ):
Sets status to 3 when worked on
"""
self._status = 3
pass
def get_disc_link( self ):
"""